The App That Knows You're Pregnant Before You Tell Your Family
The App That Knows You’re Pregnant Before You Tell Your Family
By Akshay A. Walimbe
In 2012, as reported by Charles Duhigg in The New York Times Magazine, a man walked into a Target store in the United States and demanded to speak to a manager. He was furious. His teenage daughter had been receiving coupons in the mail for baby clothes, cribs, and maternity vitamins. “She’s still in high school,” he told the manager. “Are you trying to encourage her to get pregnant?”
The manager apologised profusely. Called the man back a few days later to apologise again.
But this time, the father’s tone had changed. “I had a talk with my daughter,” he said. “It turns out there has been some activities in my house I was not completely aware of. She is due in August. I owe you an apology.”
Target’s algorithm had figured out his daughter was pregnant before she had told her own father. (Duhigg later noted he could not independently verify this specific anecdote, but the underlying programme was real and documented.)
How? A statistician named Andrew Pole had built a pregnancy prediction model for Target. By analysing purchasing patterns across thousands of customers, Pole’s team identified about 25 products that, when purchased together in specific patterns, indicated a high probability of pregnancy. Unscented lotion. Calcium and magnesium supplements. Cotton balls. Certain hand sanitisers. Individually, these are meaningless purchases. Together, they paint a picture so detailed that the algorithm could estimate not just whether someone was pregnant, but approximately when they were due.
Target assigned every customer a “pregnancy prediction score” along with an estimated due date. Then it sent coupons timed to each trimester. The system worked so well that it spooked customers. Target eventually had to disguise the pregnancy ads by mixing them in with random coupons for lawnmowers and wine glasses, so it would not look like the company was watching.
That was 2012. A single American retail chain. One algorithm. Twenty five products.
Now let me tell you about your phone.
Your Phone Knows You Better Than Your Doctor
Right now, sitting in your pocket or on your desk, your smartphone is assembling a profile of you that would make Target’s pregnancy model look like a child’s drawing.
If you live in India and you have ever ordered food, your Swiggy or Zomato app knows what you eat, when you eat, where you eat, how often you order, what your budget range is, whether you order for one person or two, and whether your patterns changed recently. A sudden shift from ordering for one to ordering for two, from late night biryani to early morning health food — that is a signal. Not just about your diet. About your life.
If you use PhonePe or Google Pay, your payment app knows your salary cycle, your rent amount, your recurring subscriptions, which medical labs you visit, which pharmacies you frequent, whether you recently started buying prenatal vitamins or baby products. It knows when your spending patterns change and in what direction.
If you use a period tracking app — and millions of Indian women do — it knows your cycle, your irregularities, your symptoms, and whether you missed a period. In 2022, Mozilla’s “Privacy Not Included” investigation found that many popular period tracking apps were sharing intimate health data with third parties, including data brokers and advertising platforms. The US Federal Trade Commission had already taken action against Flo Health in 2021 for sharing user data with Facebook and Google, despite promising to keep it private. Your most private biological information, packaged and sold.
Here is what makes this different from the Target story. Target analysed purchase data from one store. Your phone aggregates data across every app you use, every search you make, every location you visit, every payment you process. It does not need 25 products to figure out you are pregnant. It has thousands of data points, updating in real time, cross referenced across platforms.
And unlike Target, which at least sent coupons through physical mail, your phone enables companies to share or sell that inference far more rapidly, often without you knowing. To be clear, not every app is selling your data to third parties — many companies use your data only internally to improve their own services. But the point is that the architecture allows it, the incentives encourage it, and the oversight to prevent misuse is still catching up.
The Indian Data Bazaar
Let me make this concrete and local.
According to the IAMAI Kantar Internet in India 2024 report, India has over 886 million active internet users. Our digital payments ecosystem processes billions of transactions monthly through UPI alone. We have some of the cheapest mobile data in the world, which means people are online constantly — sharing, searching, transacting, living their lives through their screens.
This has created something extraordinary. Not just a digital economy, but a digital portrait of an entire nation. Every swipe, tap, search, and transaction becomes a data point. And all of it feeds into profiles that are bought, sold, analysed, and used to make decisions about you.
Consider what a single Indian fintech app can piece together. Your phone model tells them your approximate income bracket. Your PIN code tells them your neighbourhood, which in India carries deep socioeconomic signals — and, let us be honest, caste signals too. Your app usage patterns — whether you use English or Hindi, which entertainment apps you have, how often you recharge your prepaid SIM — all of this feeds into a profile that is far more intimate than anything you would share willingly.
Indian fintech companies use AI to process what the industry calls “alternative data” — text messages, app usage, device data, social media activity, expenditure patterns — to make credit decisions. As Dvara Research documented in their 2024 study on AI in digital credit in India, these AI based systems can “perpetuate biases present in data but do so at scale, without traceability.” One lending app, mPokket, openly describes processing “demographic, social, behavioral, financial and transactional data points.” They are not hiding it. It is the business model.
To be fair, these alternative data approaches also serve a purpose. With only about 600 million Indians covered by formal credit bureaus like CIBIL, alternative data is how fintech companies try to extend credit to people the traditional banking system has excluded. The problem is not the intent. The problem is the absence of transparency about what data is being used and what inferences are being drawn from it.
But here is the question nobody asks: when you downloaded that food delivery app and tapped “I Agree” to the terms and conditions, did you agree to THIS?
Did you agree to your eating patterns being analysed for health inferences? Did you agree to your payment data being cross referenced with your location data? Did you agree to a machine learning model building a prediction about your pregnancy, your illness, your financial distress, your relationship status — and then using that prediction to decide what you see, what you are offered, and what you pay?
The Consent Fiction
India’s Digital Personal Data Protection Act, 2023 — the DPDPA — requires companies to obtain your consent before processing your personal data. The law is a significant step forward. It establishes meaningful principles: purpose limitation, data minimisation, the right to erasure. Companies must now state why they are collecting your data and cannot keep it indefinitely. That matters. But on the consent question specifically, the gap between the law’s intent and everyday reality is vast.
A 2008 study by researchers Lorrie Faith Cranor and Aleecia McDonald at Carnegie Mellon calculated that the average internet user encounters roughly 1,462 privacy policies per year. Reading all of them would take 244 hours — 76 working days. A full quarter of your professional year, just to understand what you are signing away. A 2017 Deloitte survey of 2,000 US consumers found that 91 percent of people consent to terms and conditions without reading them. Among young adults aged 18 to 34, that number was 97 percent. A Pew Research Center survey in 2019 confirmed the pattern: only 9 percent of people said they always read privacy policies.
You are not consenting. You are surrendering. There is a difference.
And even if you were to read every policy, you would find that the language is designed to cover everything while explaining nothing. “We may share your data with trusted third party partners for the purpose of improving our services.” That sentence could mean anything. It could mean your food ordering data gets shared with a health insurance company. It could mean your location data gets sold to a political campaign. It could mean your financial behaviour gets fed into a credit scoring model that decides whether you get a loan.
The DPDPA says companies must tell you the purpose of data collection. But the purpose can be stated so broadly — “to improve your experience” — that it becomes meaningless. And here is the critical gap: the DPDPA does not address inference data at all. It covers the data you give a company. It does not clearly cover the conclusions a company draws from that data. The fact that you bought prenatal vitamins is personal data. The prediction that you are pregnant, derived from that purchase? That exists in a legal grey zone.
Target knew a teenager was pregnant before her father did. That was one retailer, one country, fourteen years ago.
Today, the algorithms are better. The data is deeper. The cross referencing is instant. And the inferences are being drawn not by a single company, but by an interconnected ecosystem of apps, platforms, data brokers, and AI models that share and trade information in ways that no individual can track.
So Here Is the Question
We are building a world where machines know our most intimate secrets — our health, our finances, our relationships, our vulnerabilities — before the people closest to us do. And the mechanism that is supposed to protect us, consent, was designed for a different era entirely. Consent was built for a world where someone asked you a question and you gave a clear answer. Not for a world where a tap on a screen signs away rights you did not know you had, for purposes that have not been invented yet.
You clicked “I Agree.”
But did you agree to THIS?
I’m have written a book about exactly this how AI and automated systems make decisions about your life, where accountability disappears, and what we can do about it. If you want to know morea about this book or order a copy, you can do it here: https://akshaywalimbe.com/beyond-bias/